Binary joint transform correlator thresholding on reference image only

In this work, we propose a novel thresholding method to simplify the procedure of the binary joint transform correlator. Unlike the normal binary joint transform correlator which thresholds on the power spectra of both the reference image and the target image, we threshold on the power spectra of the reference image only. With this technique the additional on-line recording of the power spectra of the target is avoided. Therefore the procedure of the recognition is as simple as that of the normal joint transform correlator. We then extend the binary joint transform correlator to circular harmonic expansion to achieve the rotation invariant optical pattern recognition. We use a computer to decompose the reference image into a sum of circular harmonic components and form an input of the joint transform correlator with the first-order circular harmonic component and the target image. By use of this technique, we achieve circular-harmonic-based binary joint transform correlation with good performance merits. Computer simulation results and 3D output correlation profiles are presented.